Visibility work has a long feedback loop with much noise. Good monitoring is mostly about patience: do not make conclusions too early.

​
A practical schedule

  • Run conversations weekly on a stable set of topics. This frequency is sufficient to find a real change, and it is not so frequent that noise hides the signal. Read How often should I run conversations?.
  • Keep the set of questions fixed between runs. If you change scenarios while you watch a trend, you measure two different things.
  • Record what you changed, and when. Almost each “why did this change” question has its answer in a note that nobody wrote.

​
How to read a change

Before you investigate a change, do these checks in sequence:
  1. Find if your competitors changed in the same direction at the same time. If the full category changed, the answer engine changed, not you. This one check prevents most of the wasted investigation.
  2. Find if the change occurs in all topics or only in some topics. A change in some topics is usually yours. A change in all topics is usually not yours.
  3. Open the conversations from before and after the change, and compare the cited sources. A changed answer almost always means a changed source.

​
What to report

Report the direction and the cause, not the number. This sentence is a result: “Visibility increased by nine points on pricing topics after we published the comparison page, and answer engines now cite it.” In contrast, “Visibility is 62%” causes a question that nobody can answer.

​
Measure what is important to you

Platform metrics describe the medium. You possibly must track a specific item for your business. For example: a recommendation for a specific phrase, in one market, against two named competitors. Use the Business Scorecard for this.